||In the last few years agent technology, especially mobile agents, became a new exciting field in computer science. Agent development is getting more and more interesting, even in commercial infrastructures, so it is worth considering in depth their strengths and the situations in which they can be used effectively. During our research work we figured out, that there exist a huge number of approaches, toolkits, and platforms of different quality and maturity. This fact led us to the problem of evaluating them to find an “optimal” (in the sense of productivity) platform.Firstly, to get an impression what should be possible for an agent platform, we collected the current research results and summarized them in a state-of-the-art report. After that, we defined a set of criteria, which collected the requirements to the platforms. Then we selected some fundamental criteria in a way that if one platform does not fulfill these criteria it is immediately knocked out from further evaluation. For the remaining platforms we applied the complete set of criteria, weighted them, and got a final evaluation result.